Search Results for author:
Found 2 papers, 0 papers with code
基于注意力的蒙古语说话人特征提取方法(Attention based Mongolian Speaker Feature Extraction)
“说话人特征提取模型提取到的说话人特征之间区分性低, 使得蒙古语声学模型无法学习到区分性信息, 导致模型无法适应不同说话人。提出一种基于注意力的说话人自适应方法, 方法引入神经图灵机进行自适应, 增加记忆模块存放说话人特征, 采用注意力机制计算记忆模块中说话人特征与当前语音说话人特征的相似权重矩阵, 通过权重矩阵重新组合成说话人特征s-vector, 进而提高说话人特征之间的区分性。在IMUT-MCT数据集上, 进行说话人特征提取方法的消融实验、模型自适应实验和案例分析。实验结果表明, 对比不同说话人特征s-vector、i-vector与d-vector, s-vector比其他两种方法的SER和WER分别降低4. 96%、1. 08%;在不同的蒙古语声学模型上进行比较, 提出的方法相对于基线均有性能提升。”
A Review of Deep Learning Based Image Super-resolution Techniques
By collecting and counting the relevant literature on the application of depth learning in the field of image super-resolution, we preliminarily summarizes the application results of depth learning method in the field of image super-resolution, and reports the latest progress of image super-resolution technology based on depth learning method.
